[HOT] [RFC] suggestions for HOT work

Oleksij Rempel bug-track at fisher-privat.net
Mon Aug 13 17:36:25 BST 2012


On 13.08.2012 15:45, Mikel Maron wrote:
> Hi
>  
>> i have some suggestions and questions (comments are welcome):
>>>
>>> - not always HOT has same kind of event, but it seems like many event
>>> have epicenteric character. For example earthquakes - we have big todo
>>> filed and different levels of damage. It will make sense if HOT website
>>> will allow to mark epicenter and manage tasks from it. It should be
>>> visible so you can decide by your self what to take, and if you press
>>> "take a task" button, you will get nearest one to epicenter.
>> I'm not sure if that's the best strategy.
>>
>> 1) An unpopulated region around the epicenter maybe is totally uninteresting.
>> 2) even a populated region that's nearer e.g. to a on-sea epicenter of an earthquake/tsunami might be less interesting if it has not been hit that hard because of a different shore shape than another region more far away.
>>
>> Perhaps I'm wrong, but that's why I don't think an automated (!) decision out of the "epicenter" information is useful.
>>
>> On the other hand it's a nice idea to have an adjustable control strategy for the task scheduling, where "start at epicenter (x,y) and take the nearest" could be one example, as well as "this is the area and the most important part for the first run is the part along that river/motorway/mountain ridge..."
>>
>> No idea how exactly that could be solved best.
>>
>>
> 
> I think there are currently two strategies implemented ... start at the top-left corner, and proceed in order; and select randomly. Starting from the center seems like another viable strategy for assigning tasks. It would be up to the admin creating the task to decide the task scheduling strategy.
> 
> I've added a ticket for this:
> https://github.com/pgiraud/OSMTM/issues/78

It will be good to have admin defined center.

>> - in many cases we need as match information as possible, but since we
>>> have not time, first thing todo are roads, power lines and residential
>>> areas. I think it will help to concentrate and coordinate this tasks
>>> better if hot-site will have multilevel tasks. Currently there is only
>>> "not ready", "ready" and "validated". How about "step one - roads",
>>> "step ... - buildings", "validated"?
>> would not set it to "steps" as some tasks may be dependent on the experience of a mapper: If I have no idea about how to "read" imagery about powerlines, I might better trace rivers and lakes first (given that roads are done), and probably on the long run there's some civil protection organisation that actively supports osm/hot mapping themselfes and there's the power coordination troup that is working on powerlines before streets are traced by mappers - but they trace it.
>>
>>
> There's a trade off in complexity here; in administering a multi-step task vs creating separate tasks for each kind of feature focus. I could also imaging integrating some statistics on each task square directly, so to show the change, comprehensive, errors/warnings around that feature.
> 

Sound good.

> 
>>> - power lines. how match are they important and if really important i
>>> would see some tool which allow creating tagged line and points at one shot.
>> I know from the THW (German Federal Agency of technical relief) that they have groups to do power supply. In the first run the set up generators for their own needs and afterwards they start to repair and secure the power infrastructure (where possible and necessary).
>>> - Currently it is second my job with lots of slums. May be some of them
>>> not really slums, but they are pain in ass if you need to draw them.
>>> Should we draw them? Or there is some kind of landuse=slum tag?
>> In general OSM (speaking not HOT specific, so there may be tagging specifically for slums, too) doesn't define mapping according to law, but according to what's on the ground.
>>
>> A slum where people live in is - on the ground, without looking into law books - some kind of residential area, as the people living there are residents, I would say.
>> Many slums on the world are de facto residential areas, even if not or not originally allowed by the authorities.
>>
>> With that I would not use landuse=slum distinct from landuse=residential.
>> If - and I'm not sure about that, I would add a secondary tag to mark that residential area as slum; but keep in mind that there's often no hard line between slum, "usual residential area" and a camp built for/of/by people hit by a desaster.
>>
>> Usually we've marked these as residential areas. I think an additional tag could be used to specify if it is an informal settlement. 

The idea is not to tag inforaml settlement. It is more about settlement
especially affected by fire and earthquake. There are many building
close/connected to each other. Take a look to current latest HOT job,
some where close to epicenter. Most small villages made in blocks. Each
block has some kind of chaotic structure. If they collapse, then there
are many people who need help.


-- 
Regards,
Oleksij



More information about the HOT mailing list